Aerodynamic Modulation Salford Uni

8/6/2019 Aerodynamic Modulation Salford Uni

1/57

Research into aerodynamic modulation of

wind turbine noise

Report by University of

Salford

URN 07/1235


2/57

Prepared for Defra byDr. Andy Moorhouse, Malcolm Hayes, Dr. Sabine von Hnerbein,Ben Piper, Dr. Mags Adams

Research into Aerodynamic Modulationof Wind Turbine Noise:Final report

July 2007Contract no NANR233


3/57

University of Salford. NANR233 Research into aerodynamic modulation of wind turbine noise

CONTENTS

1. Summary............................................................................................................3

2. Introduction........................................................................................................5

2.1. Aims and objectives...............................................................................52.2. Organisation of the project.....................................................................6

3. Survey of Local Authorities...............................................................................7

3.1. Scoping survey.......................................................................................8

3.2. Detailed survey ....................................................................................10

3.3. Survey results and analysis ..................................................................11

Number of complaints and complainants ............................................11

Prevalence of AM ................................................................................15

Significance of complaints...................................................................16

Comments on the use of complaint statistics.......................................19

Other factors arising from the survey ..................................................21

4. Further investigation of sites with AM ............................................................22

4.1. First site................................................................................................22

4.2. Second Site...........................................................................................25

4.3. Third Site .............................................................................................26

4.4. Fourth Site............................................................................................27

4.5. Conclusions from AM sites .................................................................28

5. Worldwide literature survey ............................................................................29

5.1. Search terms.........................................................................................29

5.2. Information sources .............................................................................30

5.3. Government reports and regulations....................................................31

5.4. Generation of wind turbine noise.........................................................325.5. Propagation of wind turbine noise .......................................................35

5.6. Amplitude Modulation of Wind Turbine Aerodynamic Noise............37

5.7. Psychoacoustics ...................................................................................40

6. Survey of manufacturers worldwide................................................................42

7. Conclusions......................................................................................................46

8. References........................................................................................................48

9. Appendix: Detailed survey questionnaire........................................................54

Page 2 of 57


4/57


1. Summary

The study described in this report has been commissioned by Defra, BERR (formerlyDTI) and CLG. It follows on from a report by the Hayes McKenzie Partnership to

DTI in 2005 in which reports of low frequency noise emission from windfarms were

investigated. Their report concluded that the complaints were not caused by low

frequency noise, but by amplitude modulation of aerodynamic noise (AM) from the

wind turbines. The term AM indicates aerodynamic noise from wind turbines, but

with a greater than normal degree of regular fluctuation at blade passing frequency,

typically once per second. The aims of this current study are to ascertain the

prevalence of AM on UK wind farm sites, to try to gain a better understanding of the

likely causes, and to establish whether further research into AM is required. The study

was carried out in four parts, a survey of local authorities with windfarms in their

areas, further investigation of sites for which AM was identified as a factor, a

literature review and a survey of wind turbine manufacturers.

The survey of local authorities was in two parts, a scoping survey aimed at identifying

problem sites, and a detailed survey to establish whether AM could have been a factor

in causing complaints. The response to both parts of the survey was 100%, although

full information was not available for all sites at the detailed stage. The results showed

that 27 of the 133 windfarm sites operational across the UK at the time of the survey

had attracted noise complaints at some point. An estimated total of 239 formal

complaints have been received about UK windfarm sites since 1991, 152 of which

were from a single site. The estimated total number of complainants is 81 over the

same sixteen year period. This shows that in terms of the number of people affected,

wind farm noise is a small-scale problem compared with other types of noise; for

example the number of complaints about industrial noise exceeds those about

windfarms by around three orders of magnitude. In only one case was the windfarm

considered by the local authority to be causing a statutory nuisance. Again, this

indicates that, despite press articles to the contrary, the incidence of windfarm noise

and AM in the UK is low.

Page 3 of 57


5/57


AM was considered to be a factor in four of the sites, and a possible factor in another

eight. Regarding the four sites, analysis of meteorological data suggests that the

conditions for AM would prevail between about 7% and 15% of the time. AM would

not therefore be present most days, although it could occur for several days running

over some periods. Complaints have subsided for three out of these four sites, in one

case as a result of remedial treatment in the form of a wind turbine control system. In

the remaining case, which is a recent installation, investigations are ongoing.

The literature review indicated that, although there has been much research into the

general area of aerodynamic noise it is a highly complex field, and whilst general

principles are understood there are still unanswered questions. Regarding the specific

phenomenon of AM there has been little research and the causes are still the subject of

debate. AM is not fully predictable at current state of the art. The survey of wind

turbine manufacturers revealed that, although there was considerable interest, few

have any experience of AM.

The low incidence of AM and the low numbers of people adversely affected make it

difficult to justify further research funding in preference to other more widespread

noise issues. On the other hand, since AM cannot be fully predicted at present, and its

causes are not fully understood we consider that it might be prudent to carry out

further research to improve understanding in this area.

Page 4 of 57


6/57


2. Introduction

In 2005 the DTI commissioned the Hayes McKenzie Partnership to investigate

claims in the press that infrasound or low frequency noise emitted by wind farms

were causing health effects. Their report, published in 2006, concluded that there

was no evidence of health effects arising from infrasound or low frequency noise

generated by wind turbines. The report went on to note that the cause of

complaints was not low frequency noise or infrasound, but was audible

modulation of aerodynamic noise, i.e. aerodynamic noise which displays a greater

degree of fluctuation than usual. This phenomenon is referred to as AM in the rest

of this report. This AM was, in some isolated circumstances, occurring in ways

not anticipated by the government guidance document on noise from windfarms

known as ETSU-R-97. The Government therefore took the view that more work

was required to determine whether or not AM is an issue which may require

attention in the context of the rating advice given in the ETSU guide.

2.1. Aims and objectives

The aims of this study are to ascertain the prevalence of AM from UK wind farm

sites, to try to gain a better understanding of the likely causes, and to establish

whether further research into AM is required.

The objectives of the study are as follows:

(a) To establish the levels and nature of the reported noise complaints received

across the UK relating to noise issues from wind farms, both historic and current,

and determine whether AM is a significant effect;

(b) To review and understand the level of knowledge/understanding that exists

throughout the world on AM, and whether AM can be predicted;

Page 5 of 57


7/57


2.2. Organisation of the project

The study has been organised into four main tasks in order to achieve these

objectives:

a. Survey of Local Authorities;

b. Further investigation into sites with AM

c. Worldwide literature search;

d. Survey of Manufacturers Worldwide;

Sections 3-6 of this report describe tasks a-d respectively, followed by overallConclusions in Section 7.

Page 6 of 57


8/57


3. Survey of Local Authorities

A two-stage survey of local authorities was conducted with the aim of establishing the

presence of noise complaints, in particular about AM, from UK windfarms. A

database of all operational wind farms in the UK was provided to us by BWEA. This

gave us a total of 136 wind farms that were operational at the end of December 2006

to investigate. Four of these are offshore sites, which were initially excluded from the

survey on the grounds that they are far from habitation, but were later included after it

emerged that one of them had in fact received complaints (which later turned out to be

from foghorns, not the wind turbines). The BWEA list includes some windfarms, that

are extensions of existing sites, as separate developments. Such extensions are not

considered as separate sites by the local authority, and so were merged together on the

database. The total number of separate windfarm sites then becomes 133, of which 4

are offshore. Smaller wind energy developments, for example single wind turbines,

were not included in the survey because there is no is no definitive list of such

developments and it would therefore be difficult to ensure we were dealing with a

representative sample.

The BWEA database also provided information about the number of wind turbines in

each wind farm, their capacity, the name of the developer, the country and county of

location within the UK and the LA responsible for giving planning permission.

We confirmed that noise nuisance is dealt with by either Unitary Authorities or

District Authorities and then identified specific contacts within each local authority.

It was felt that where possible we should identify a named individual within the local

authority rather than sending a general enquiry to the head of department or enquiry

email address. Using the membership list of the Institute of Acoustics we identified

names, email addresses and telephone numbers of Environmental Health Officers

within each local authority with expertise in noise and acoustics. We also used the

Noise Abatement Societys on-line database of local authority contacts for noise

complaints to identify contact details for local authorities which did not have

Page 7 of 57


9/57


individual members of the IoA. We also used a database of Local Authorities held at

the University of Salford. If no contact was found we directed our enquiries to the

head of the Pollution Control Section or Environmental Health. Any missing contacts

were identified by phone calls to the local authority.

On a map we identified all wind farms that were on local authority borders so that we

could check if any noise complaints were received by neighbouring local authorities.

47 windfarms were identified as being close to the border with a neighbouring

authority, and these authorities were added to the database of local authorities to be

contacted.

To respect confidentiality the names of individual windfarms will not be used in this

report, nor will the names of specific local authorities or their members of staff. Each

windfarm will be referred to by a letter or letters of the alphabet.

3.1. Scoping survey

The first stage of the survey, a scoping survey, aimed to identify those local

authorities who had received complaints about windfarm noise. The questionnaire

was sent out by email to named individuals and was designed to be as simple as

possible to answer with a simple Yes/No, so as to maximise the chances of a reply.

Several points on the questionnaire required careful consideration. One of the main

decisions was the question of how far back to go with complaints. It is known that

some early installations produced complaints about tonal noise from wind turbine

nacelles, which are not related to AM. After some discussion it was decided that to

specify a start date might weaken any conclusions that could be drawn. On the other

hand, not to specify any date could lead to inconsistencies in the responses. It was

therefore decided to specify that complaints should be included from any time since

the windfarms have been operational.

A copy of the final phase 1 questionnaire is shown in Figure 1.

Page 8 of 57


10/57


Dear NAME

We at Salford University in conjunction with Hayes McKenzie have been appointed by the UK

Government (funded by Defra, DTi, and DCLG) to undertake research into noise from

windfarms. For details of why this work was commissioned please see

http://www.dti.gov.uk/files/file35043.pdf . The aim of the work is to provide clearer advice to

planning authorities about wind turbine planning applications.

We would appreciate it if you could take a couple of minutes to answer the following

question.

Have you had any noise complaints about any of the wind farms in your area or any

windfarms in a bordering authorityany time since they have been operational?

If the answer is NO please respond by pressing REPLY in your browser window and simply

type NO. We will not trouble you further.

If the answer is YES please respond by pressing REPLY in your browser window and simply

list the names of the windfarms about which there have been noise complaints. If the answer

is yes then we will need to follow up with a further survey in which we will ask for further

details.

We appreciate your help with this matter as the Government wants a 100% response to this

question and would be grateful if you could reply by DATE.

Sincerely, Anne Marie Lavin

Sent on behalf of

Head of Survey of Local Authorities

Government Survey into Noise from Wind Turbines (REF: NANR233)

Dr Mags Adams

Senior Research Fellow

Acoustics Research Centre

University of Salford M5 4WT

Figure 1 Phase 1 scoping survey questionnaire

Page 9 of 57
http://www.dti.gov.uk/files/file35043.pdfhttp://www.dti.gov.uk/files/file35043.pdf


11/57


After some follow up emails and, where needed, telephone reminders we were

successful in achieving a 100% response to the scoping survey.

3.2. Detailed survey

The aim of the detailed survey was to obtain information about the number and nature

of complaints, and in particular to establish the contribution of AM. Every local

authority that responded with a positive noise complaint in the scoping survey was

sent the detailed questionnaire, to be completed for each wind farm in their area for

which they had received noise complaints. In total, 23 LAs were contacted about 27

windfarms.

The detailed questionnaire was developed by the team at Salford, was reviewed by

EHOs on the Noise Working Group and by the Hayes McKenzie Partnership, and

then sent out via email to the Local Authorities. A complete copy of the final

questionnaire is given in the Appendix. The questions include a combination of open

and closed questions with tick-boxes to simplify responses where possible. It was

decided not to mention AM until the final question so as to encourage EHOs to

provide information about all noise complaints and not just to focus on those that

might fit the description of AM problems.

Details about the number of complaints and number of complainants were requested

along with detailed descriptions of the noise itself and the investigation that took

place. A list of terms to describe noise from wind turbines was provided and the

EHOs were requested to indicate which of them corresponded to the complaints they

had received from each wind farm. Details of any action that had been taken was also

requested. Finally, the EHO was asked to indicate whether they felt that the

complaint conformed to a description of Amplitude Modulation of Aerodynamic

Noise.

Local Authorities were asked to respond within a week and were sent a reminder

email and follow up phone calls as necessary.

Page 10 of 57


12/57


3.3. Survey results and analysis

We obtained a 100% response rate from both the scoping and detailed questionnaires.

However, complete information for the detailed survey was not always available. This

was particularly the case for complaints from the 1990s since many local authorities

did not keep computer records at that time.

Number of complaints and complainants

From the results of the scoping survey we determined that 27 out of the 133

windfarms operating in the UK at the time of the survey have attracted formal

complaints about noise. This represents 20% of the total, i.e. one in five windfarms

has attracted complaints about noise at some point. One other site, an offshore site,

received complaints about foghorns, and this case has been excluded in the statistics

since the wind turbines themselves did not generate the noise. These figures are based

on 100% returns from local authorities throughout the UK over the entire period since

the first windfarms became operational in 1991, and can therefore be treated as a

comprehensive and definitive picture of formal noise complaints about windfarms in

the UK.

Complete results obtained for the numbers of complaints, numbers of complainants,

and the years in which complaints were received are shown in Table 1. Note the

distinction between complainant and complaint, where complainant refers to the

individual who makes the complaint and complaint refers to each time they log a

complaint with the local authority. One complainant may make a number of

complaints and therefore the number of complaints can exceed the number of

complainants. In order to provide some context for the complaint histories the bottom

row ofTable 1 gives the number of sites on-line at the end of each year.

We have specific information on the number of complaints from 19 of the 27

windfarms. The total number of complaints received by local authorities relating to

these 19 sites was 206, 152 of which were received for a single windfarm. By

Page 11 of 57


13/57


Page 12 of 57

excluding this untypical figure we see the number of noise complaints about each

windfarm varied between 1 and 10 with an average of 3. We can use this to estimate

the total number of complaints across the UK by assuming that this average number

also applies to the eight sites for which no specific complaint data was available. The

result of this projection is that the total number of complaints across the UK is

approximately 230 since records began, 152 of which were from a single site. These

statistics about the number of complaints should be treated with some caution because

there is not complete consistency in the way local authorities count complaints: some

would count several complaints from the same person as a single complaint, others

would log each separate complaint individually. For this reason it is probably more

reliable to count the number of premises affected, i.e. the number ofcomplainants as

described in the next paragraph.

We now consider the results for the number of complainants. The number of

complainants will generally be less than the number of complaints because the same

complainant may make several complaints. We have firm information about the

number of complainants from 16 sites, with a total of 53 complainants (these are the

entries in the rightmost column ofTable 1 without an asterisk). No complainant data

was available for the remaining 11 sites. For sites marked with an asterisk in the

rightmost column ofTable 1 the number of complainants was estimated. We did this

by making the conservative estimate that each complaint was made by a different

complainant. These estimates may be slightly on the high side, but since in every case

the numbers are low we would not expect this to have a large influence on the totals.

Using this approximation we estimate the number of complainants relating to the 25

windfarms for which we have either complainant or complaint data to be 75. The

number of complainants per windfarm varied from 1 to 9 with an average of 3. As

before, we can project the total number of complainants across the UK by assuming

this average applies to each of the two sites for which no definitive information is

available. The final result is that an estimated total of 81 households have complained

about windfarms in the UK since records began in 1991.


14/57


Complaints per year and total complaints

Windfarm

Total

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

Yearof

complaints

A 3 2 1 2005

B 10 10 2005

C 1 1 1994

D 152 1 7 10 9 52 50 23 2001

E 4 3 1 2003

F 2 1 1 1996,1998

G 4 1 1 1 1 2000, 2002, 2005

H 1 2005

I before 2003

J 2 2 2005

K 7 7 2006

L 2 2 2004

M 4 1 1 1 1 1997,1998,2000,2

N 2 1 1 2002, 2006

O 1 1 2007

P 1993,1994,1995,1

Q 2001, 2002, 2003

R 1992,1993S before 1997

T

U 1 1 2006

V 1 1 2004

W 2 2 2007

X 2 2 2006

Y

Z 2 1 1 2000, 2001

AA 4 1 3 2005, 2006

No of entries 19 No of entries

No of windfarms 27 No of windfarms

% coverage 70% % coverage

Total complaints 206 4 23 32 12 53 52 24 2 0 2 1 1 0 1 0 0 Total complainant

Total less D 54 3 16 16 3 1 2 1 2 0 2 1 1 0 1 0 0 Average per wind

Ave (less D) 3.0 Total projectedTotal projected 230 *No separate information given on number of complainants, assumed same as number of c

Total proj. less A 78

136 114 95 83 75 64 56 45 40 36 31 24 18 14 7Sites on line at end of year

Table 1 Summary of complaint and complainant history data. Shading indicates the years in which complaints occurred

known. Blank entries indicate no data was available.

Page 13 of 57


15/57

versity of Salford. NANR233 Research into aerodynamic modulation of wind turbine noise

Page 14 of 57

Windfarm

trainnevergets

there

Distant

Helicopter

Thumping

Thuddng

Pulsating

Rythmatical

Beat

Beating

Throbbing

Lapping

Wooshing

Swish

Swoosh

Ghostlynoises

WoohWooh

Grinding

Rumbling

likemotion

sickness'

Whistling

Otherplease

listOther(describe)

A Y Y Y Y Y Y aircraft landing, dist

B Y Y Y Y Y Y YC

D Y Y Y Y Y Y Y Y Y Y train in next field, pe

E Y Y Y Y churning head

F

G

H

I

J Y Y Y Y Y Y Y mechanical

K Y Y Y Y Y Y aeroplane overhead

L Y Y low frequency droni

M Y Y Y Y Y Y Y Y

N w ne

O Y Y Y Y Y Y Y

P

Q

R

S

T

U Y Y Mechanical whine a

V Y

W

X Y Y Y Y Y Y Y Y Y Y Y Hum

Y

Z Y Y Y Y Y

AA Y like a washing mach

Table 2 Summary of detailed survey returns and analysis to determine the prevalence o

Uni


16/57


Prevalence of AM

In this section we consider the extent to which AM may have been a factor in the

complaints. Summarised in Table 2 are the results from the part of the survey where

local authorities categorised the noise associated with each site based on

complainants descriptions. Also given in this table are columns indicating whether

the investigating environmental health officer (EHO) heard the noise, whether they

considered it a nuisance and whether they considered the noise to be consistent with

the description of AM.

In order to determine whether AM was a factor for each site we took into account

several pieces of information: the tick box descriptions of the noise, any other

description of the noise, general descriptions of the EHOs investigation, the timing of

the complaints, and in some cases a personal knowledge of the site in question. We

also considered the EHOs opinion as to whether the noise was consistent with AM,

and this information was assigned more confidence if they had heard the noise

themselves.

Columns 2-20 are the results from the tick box questions on the survey. The columns

have been arranged so that those on the left are the most likely to indicate AM, i.e.

like a train that never gets there, distant helicopter, thumping, thudding,

pulsating, thumping, rhythmical beat, and beating could be indicative of AM.

The terms on the right are generally associated with some other noise source such as

gear box noise or blade resonance.

For several sites, particularly early ones, the complaint was cause by a mechanical

fault or a gearbox and therefore AM can be excluded. Five sites (F, G, M, N and T)

were clearly stated in the returns to be in this category, and four others were possible

or probable (P, R, S and Y) on the grounds that they were early problems that have

not attracted complaints in recent years. AM was therefore not considered to be the

cause of these complaints, the only exception being site M where there were two typesof complaint, one indicating mechanical sources and the other indicating possible

Page 15 of 57


17/57


AM1. From the tick-box descriptions we also concluded that sites L and U were

unlikely to involve AM. The other descriptions given by complainants further

suggested that mechanical noise was a factor for sites J, L, N, and U. Altogether, AM

could be excluded as a factor from eleven sites on the grounds that the complaints

were about mechanical noise sources.

The penultimate column of Table 2 indicates that of the 27 sites considered, eight

EHOs said AM was not a factor, four said it was, ten did not know whether it was a

factor or not and five offered no opinion. The final column in Table 2 gives our

opinion as to whether the complaints could be associated with AM taking into account

all the information available. In all cases, except one, where the EHO had given a

clear Yes or No answer to whether AM was a feature our opinion was consistent with

theirs. The one exception was for windfarm AA where the EHOs opinion was No

but where we felt that some of the descriptions could have indicated AM. We

therefore entered this site as a Maybe.

The final tally is that AM is thought to be a factor in four cases, is not a factor in

fourteen, and eight cases are marginal. One is too recent to have sufficient information

to form an opinion and has been entered as Dont know on Table 2.

Significance of complaints

In this section we discuss the significance of the complaints by comparing statistics

for windfarms with similar statistics for other noise sources. There are two possible

ways in which to make comparison, in terms of absolute numbers of complaints, or in

terms of the relative number of installations attracting complaints.

Considering a comparison in terms of absolute numbers of complaints, the Chartered

Institute of Environmental Health (CIEH) publishes statistics annually on the number

1

This site is the only site in the UK still to employ outdated two-bladed rotors, which are known tocause greater modulation than more modern three-bladed machines. AM is therefore a potential

explanation of complaints in this case, but this does not cause concern about the future.

Page 16 of 57


18/57


of noise complaints and nuisances throughout England and Wales. A summary of

some of the results from 2004/2005 is given in Table 3.

The number of households complaining to local authorities about noise from

windfarm sites throughout the UK since 1991 has been calculated in the preceding

section as 81, which corresponds to an average of approximately 5 per year. The

number of complaints was projected as 230, an average of about 14 per year, although

we have less confidence in this figure than that for the number of complainants. From

Table 3 the number of complaints about industrial noise (probably the most

comparable category with windfarms at least in terms of the character of the noise)

was 7,522 in a year. The total number of complaints about noise was 286,872. It

should be noted that the figures from Table 3 are based on returns from 69% of local

authorities in England and Wales, whereas the figures for windfarms are from 100%

of local authorities throughout the UK (i.e. also including Scotland and Northern

Ireland), so that the comparison will tend to overstate the significance of windfarm

noise in relative terms.

Industrial

Commercial/

leisure

Domestic

Construction/

demolition

Vehiclesmachinery

andequipmentin

streets

Miscellaneous

Total

Noise incidents complained of 7.5 35.8 206.1 11.7 11.3 14.5 286.9

Nuisances 1.4 6.0 27.6 1.8 1.7 1.0 39.5

Table 3 Summary of CIEH statistics on noise complaints and confirmed nuisances (thousands)

from 2004/2005 for 69% of local authorities in England and Wales.

It is clear that complaints about noise from windfarms make up an extremely small

proportion of the total noise complaints: complaints about industrial noise exceed

those from windfarms by around three orders of magnitude, and complaints about

noise in general exceed those from windfarms by between four and five orders of

magnitude. We would stress that this does not imply that individual complaints about

windfarms are less important than about other noise sources, but rather that the scale

Page 17 of 57


19/57


of the problem in absolute terms is significantly smaller than for other categories of

noise.

Clearly a major factor in the small number of complaints relating to windfarms is the

relatively small number of sites compared to, say, industrial sites in general. It would

be interesting to make a comparison in terms of the proportion of sites of a particular

type that attract complaints. However, whilst we know from this study that about one

in five windfarms attracts complaints at some point, we do not have comparable

figures for other types of site and we are not therefore able to compare in relative

terms.

We can gain further insight into the significance of windfarm noise by looking at

nuisance statistics. We note from Table 2 that for 16 windfarm sites the local authority

considered the noise from the windfarm not to be causing a nuisance, four cases were

pending at the time of the survey, i.e. the local authority had not yet reached a formal

decision, and in only one case was the noise judged to be causing a statutory nuisance.

There are several categories of nuisance in law, and in order for a noise to be

actionable by a noise abatement notice it must be categorised as a statutory nuisance

a formal definition of which is given in the Environmental Protection Act 1990. We

confirmed that the one Yes site had been categorised by the local authority as a

statutory nuisance. For the 16 sites not considered to be causing a nuisance it can

be taken that there was also no statutory nuisance. Therefore, there is only one

confirmed case of statutory nuisance from the 17 sites for which data is available.

Note again that these figures cover a 16 year period. In comparison, Table 3 shows

that there were a total 39,508 confirmed noise nuisances in 2004/2005, of which 1,401

related to industrial noise. Again, it is clear that windfarm noise is an extremely small-

scale problem compared with other types of noise, particularly since the figures for

other sources are an underestimate for the UK as a whole as remarked earlier.

One further observation from the nuisance statistics is that the proportion of

complaints which were subsequently confirmed as causing a nuisance is smaller for

windfarms than for any of the other categories of noise as used by the CIEH and given

in Table 3. The proportion for industrial noise was about 19% in 2004/2005, and for

noise complaints in general it was about 14%. The comparable figure for windfarms is

Page 18 of 57


20/57


about 6% based on one nuisance per 17 cases for which we have firm data from Table

2. These figures could be interpreted in several ways, although one should be aware

that the number of results is too small to draw firm conclusions, and also that the

picture could change if some of the four pending cases is eventually judged a

nuisance. One possible interpretation is that expectations are higher for windfarm

noise than for other categories of noise. If so, we might speculate that the newness of

windfarms as a noise source might be a relevant factor. Another possible

interpretation might be that local authorities are somehow more lenient for windfarms

than for other categories of noise sources, although this seems unlikely since the role

of an Environmental Health Officers is to act as an independent judge. We might also

speculate that Environmental Health Officers may not have the expertise or

confidence in their experience to challenge an ETSU report and are therefore more

reluctant to declare a nuisance compared with less complex and better understood

noise sources. Again, the fact that windfarm noise is relatively recent phenomenon

could be a factor here. Several other factors might influence the decision of the

investigating officer, but these are not particular to windfarms so will not be discussed

here as they would not explain differences in the proportion of nuisances between

categories.

Comments on the use of complaint statistics

One should, of course, be cautious about placing too much reliance on data derived

purely from reported complaints. The first reason is that not everyone who is

adversely affected by a noise goes to the trouble of making a formal complaint. Thus,

complaint statistics tend to underestimate the number of people adversely affected.

This is a recognised effect, and applies to all noise sources, but to different extents.

For example there is evidence that people tend to complain less about traffic noise

than other noise sources for the same degree of annoyance because they do not feel

that it will have any effect. It seems likely that, if anything, people might be more

inclined to complain about windfarms than about other noise sources because of the

high level of publicity surrounding the issue, but we have no definitive information on

this question. Therefore, we have no basis for assuming that windfarm noise is anydifferent to, say, industrial noise in this respect.

Page 19 of 57


21/57


The second reason for caution is that there is no control on complaints: anyone may

make a complaint whether or not they have reasonable cause for dissatisfaction.

People can, and frequently do, use noise as a proxy for some other grievance, and thus

a proportion of noise complaints may be motivated by some other factor entirely.

Thus, the fact that a site receives complaints does not imply that the noise situation is

unsatisfactory. Conversely, a lack of complaints is no guarantee of a satisfactory

situation. Although we did not ask specifically about other motivations in the

questionnaire, some local authorities volunteered the view that ulterior motives were a

factor, although not necessarily the only factor, in the complaints they had received.

This was the case for three sites, and in one of these it was felt to be a strong factor.

The same conclusion may apply to other sites but we have no definitive information.

The third reason for caution is that some complaints may go direct to windfarm

operators without being recorded by local authorities. Many local authorities now

require, as part of planning permission, the operators to undertake noise monitoring

and will forward details of complaints to the operator for further investigation. If

complainants were to go directly to the operator rather than via the local authority

then some complaints may not have been recorded by the local authorities. There is no

legal requirement on the part of the operator to report noise complaints that are made

directly to them to the local authority.

In order to try to estimate the scale of this effect we carried out a telephone follow-up

to the detailed survey, in which we asked four local authorities, between them

responsible for 10 of the 27 complained-about windfarms, if they thought it likely that

they had missed many complaints. The answers were consistent. Firstly, none of the

local authorities had formal procedures by which the operator was required to forward

details of any complaints to them, and therefore they could not guarantee that they had

records of all complaints. However, all four local authorities also felt it unlikely that

many, if any, complainants had escaped their records entirely. In some cases they felt

that local residents would not know who the operator was, so would complain first to

the local environmental health department. Some felt that since they were in close and

open contact with developers to try to resolve known problems that they would have

become aware of additional complaints. Others commented that they specifically

Page 20 of 57


22/57


requested complainants to contact them again if the problem was not resolved to their

satisfaction.

Therefore, whilst it is possible that there have been some noise problems that are not

reflected in the results of the detailed survey, we consider it unlikely that significant

numbers of complainants have escaped local authority records. In any case, windfarm

noise is no different to other types of noise in this respect: for example, a complaint

about industrial noise may be directed to the operator of the site and may never

feature on local authority records, particularly if the problem is resolved. Therefore,

the comparisons with other noise sources in the previous section should be reliable.

Other factors arising from the survey

Some other possible causes of noise complaints emerged as part of this exercise.

Unrealistic expectations was thought to be a factor in at least one case, i.e. the

complainant believed that the noise would be less noticeable than it actually proved to

be. In this case the local authority considered that the complaint could possibly have

been avoided by more accurate information from the developer at the planning stage.

Another possible cause is that complaints are being prompted by planning

applications to build other sites nearby. This was mentioned as a factor in two or three

cases. In one further case the sheltered location of the complainants property was

identified as contributing to the complaint, i.e. the background noise at the property,

which normally helps to mask noise from the windfarm, does not increase when

conditions are windy at the wind farm.

Finally, for five of the sites having received complaints EHOs commented that the site

met planning conditions. In one of these cases AM was a factor, in three it was not,

and in the fifth it was possible. We do not know whether these comments indicate that

complaints were unjustified, or alternatively that planning conditions were too lenient.

Page 21 of 57


23/57


4. Further investigation of sites with AM

In the previous section the prevalence of AM in terms of the number of sites affected

was determined. In this section we evaluate in more detail the four sites for which AM

is thought to contribute to complaints, with the aim of obtaining an estimate of the

proportion of time for which AM might occur. This estimate has to be based on

complaint data, since it is beyond the scope of this project to take measurements.

Using the complaint data we can carry out the following steps:

a. obtain complaint logs;

b. determine the range of wind speed and direction prevailing at times when

complaints occur;

c. obtain continuous records of wind speed and direction occurring at the site;

d. calculate, by comparing b. and c., the proportion of the time for which the

wind conditions associated with complaints prevail.

A second aim of this section is to determine whether complaints at these sites are

historic or ongoing.

4.1. First site

For the first site, a complaints log was available giving date and time of complaints,

together with wind speed and direction. Within this five month period complaints

were registered for 61 hours out of 2510. Hourly averaged wind data from the MIDAS

Weather Station nearby were also obtained. Both the complaints and MIDAS data are

plotted on Figure 2. The high degree of agreement between the MIDAS and complaint

log data gives confidence that the data from the Weather Station are representative for

the wind farm site. Figure 3 gives the same wind direction data, plotted in a different

format: the distance from the centre of the plot indicates the proportion of the time for

which the wind was from the given direction. The purple line is from the weather

station data and shows that over the whole period the prevailing winds were fairly

evenly distributed between 150 and 340. The blue line relates to the complaints and

indicates that most complaints occur for directions around 200.

Page 22 of 57


24/57


0 20 40 60 80 100 1200

2

4

6

8

10

12

14

16

18

Time [days 17/8-27/11/2001]

Windspeed[m/s]

complaints statistics

MIDAS Weather Station

0 20 40 60 80 100 1200

50

100

150

200

250

300

350

400

Time [days 17/8-27/11/2001]

Windd

irection[deg]

complaints statistics

MIDAS Weather Station

Figure 2 Time series of wind speed (top) and direction (bottom) for MIDAS data (green line) and

data from the complaints log (blue circles) at first site.

5 1015 20

30

210

60

240

90270

120

300

150

330

180

0

Figure 3 Distribution of hourly wind direction averages at first site: % for complaints (blue solid

line); % for weather station data during the whole analysis period from Figure 2 (black dotted

line).

From the data in Figure 2 and Figure 3 we can now try to give a figure for the

proportion of time for which wind conditions associated with complaints are likely to

Page 23 of 57


25/57


arise. First, we note that complaints only occurred for wind speeds exceeding 5m/s,

which is the cut-in speed of the turbines. Second, all complaints occurred for wind

directions in the range 140 to 270, i.e. for a 130 arc encompassing southerly and

westerly directions. However, looking further at Figure 2 and Figure 3 we see that all

parts of this range are not equally important, with the majority of complaints (73%)

occurring within a narrower 40 arc between 180 and 220 (as shown by prominent

lobe in Figure 3). Wind directions within this 40 arc were consistently associated

with complaints whereas angles of between 220 and 270 were sometimes, but not

always, associated with complaints. Furthermore, when complaints occurred at the

higher values of wind direction the descriptions from the complaint logs are generally

less suggestive of AM as a cause. Thus, the 130 arc is probably conservative because

it includes conditions which do not consistently yield complaints and the complaints

do not tend to suggest AM. On the other hand, the 40 arc includes most but not all

complaints so might be slightly unconservative.

The weather station data has been analysed taking the wind conditions characteristic

of complaints to be: speed above 5m/s and direction 140 to 270 i.e. the 130 arc.

These meteorological conditions were met in 580 hours out of the 2510 hours of

monitoring, which corresponds to 23 % of the time. Taking the smaller 40 arc as

characteristic of complaints, the corresponding figure is 5.3%. As weather conditions

change with the seasons and sometimes even from year to year a period of several

years (1/1/2000-1/1/2007), was analysed for the Weather Station. The wind direction

data for this period are shown in Figure 4 and Figure 5. The identified meteorological

conditions during this period were met for 25% of the time taking the 130 arc, and

9% taking the 40 arc. If we assume that the complaints were caused exclusively by

AM then we can conclude that meteorological conditions likely to cause AM occur

for more than 9% and less than 25% of the time at this site, with the true value

probably towards the lower end of this range, say about 15% as an overall figure. We

can therefore say that the conditions associated with AM would occur on average

about one day in seven. However, it should be noted that these are average figures and

that in reality the characteristic conditions might occur quite irregularly. This is shown

in Figure 5 which plots the wind speed for periods when the wind direction was

within the 130 arc as mentioned above. It is seen that the characteristic conditions

Page 24 of 57


26/57


might occur for several days running followed by a longer period with no such

conditions.

10002000

3000

300

2100

600

2400

9002700

1200

3000

1500

3300

1800

0o

Figure 4 Distribution of hourly wind direction averages averaged over a seven year period for

first site.

Figure 5. Wind speed for the periods when AM might occur at first site, i.e. wind speed >5m/s

and direction between 140 and 270.

The final question about this site concerns whether AM is historical or ongoing. The

Local Authority for this site has reported that there have been no complaints since

2004, suggesting either that the problem has been alleviated, or that people have

become accustomed to it, or a combination of both.

4.2. Second Site

For the second site a complaints log was available prior to 2003. This site was the

subject of complaints associated with a number of wind directions, however, it was

generally agreed that AM occurred specifically for Easterly winds and for speeds

from the cut-in speed, of around 5m/s, up to 10 m/s measured at a height of 10m

above ground level. Above 10m/s (which corresponds to about 13m/s at hub height)

Page 25 of 57


27/57


the noise of wind in the trees would start to mask the sound of the wind farm. No

wind data was available for this site; generally in the UK easterly winds are relatively

uncommon suggesting that the conditions for AM would not be present most days of

the year, although there could be periods of continuous easterly winds for several days

at a time.

AM on this site was associated with three specific wind turbines. To alleviate the

problem, a turbine control system was programmed to shut down these three machines

for wind directions between 55 and 130, and for wind speeds from the cut-in speed

to a hub height wind speed of 13 m/s. The benefits of this system may be seen in the

reduction of complaints which reduced from 50 52 per year for 2002 2003 to 7

10 per year between 2004 and 2006. The residual complaints may be associated with

westerly winds rather than the easterly wind which resulted in the audible AM. We

might cautiously conclude that the AM problem on this site has been solved, or at

least substantially alleviated by these measures.

4.3. Third Site

For the third site a complaints log was available for the period when the wind farmcommenced operation to date. The log details a number of sounds that are audible at

the property. The complainant describes periods of operation when amplitude

modulation of the aerodynamic noise (AM) is clearly audible inside and outside the

building. When the AM is described, the common factor between different days is the

wind direction, which falls into a narrow range between 140 and 170, i.e. a SSE

wind tending to a Southerly wind direction.

Analysis of the historical wind rose (Figure 6), which was supplied by the wind farm

operator and which is based upon the 10 year average data supplied by the

Meteorological Office for a nearby reference site and the on-site wind speed

measurements undertaken by the developer for a one year period, indicate that this

range of wind directions would be expected to occur for 7 9% of the year. This does

not account for the range of wind speed likely to cause AM, i.e. at very low wind

speeds the turbines would not be operating and at very high wind speeds the natural

background noise would tend to mask wind turbine noise. With periods of very high

Page 26 of 57


28/57


and low wind speed removed, the average year will see the wind from this direction

for 7% of the time. We therefore conclude that the conditions likely to cause AM on

this site will be present for approximately 7% of the time. It should be mentioned that

other noise issues are also reported at this site, but these are not considered in this

report which focuses on the issue of AM.

Figure 6 Wind farm rose for third site

4.4. Fourth Site

No complaints logs were available from the fourth site from which to ascertain the

characteristic meteorological conditions, and therefore the prevalence of AM could

not be determined. Regarding the question of whether the problem has been solved,

complaint history suggests that the problem is not a continuing one since the last

complaint relating to AM was in 1998. There have been other complaints since then,

but relating to specific fault conditions (not associated with AM) which were rectified.

According to the Local Authority, AM probably does occur at this site, i.e. there is

significant modulation of the aerodynamic noise from the windfarm, but the noise

Page 27 of 57


29/57


levels are not sufficient at nearby properties to cause a significant problem.

Furthermore, the site is operating within its planning conditions, and in this respect

the Local Authority does not consider the site to be a continuing problem. Other than

regular noise monitoring the operator has not made any remedial works to the site that

would have caused the complaints to subside. Therefore, it is possible that the initial

complaints, which occurred when the site was still new, were due to the unfamiliarity

of the noise.

To conclude, whilst it appears that AM does occur at this site, it also appears to be at

most a marginal problem.

4.5. Conclusions from AM sites

To summarise, the meteorological conditions associated with AM have been assessed

for two sites, and are likely to occur between 7% and 25% of the time, although the

latter figure is on the conservative side. Of the four sites where AM was judged by the

Local Authority to be present in only one case, which is a recent installation, is there

an ongoing investigation. In two cases complaints have largely or completely

subsided, in the final case a turbine control system has been installed and, althoughthere are some ongoing complaints, AM is probably not the cause.

Page 28 of 57


30/57


5. Worldwide literature survey

The aim of the literature survey is to summarise the current scientific knowledge

about AM and immediately related subjects. Our main information sources were peer-

reviewed documents, that is reliable, independently checked sources, such as books,

government reports and journals. Where necessary, proceedings from scientific

conferences are also considered. Under the References headline we list all the

publications we have found and read but not necessarily explicitly cited in the text.

The subjects in the literature review cover how AM is accounted for in government

guidelines and regulations, how noise in general and AM in particular are generated

by wind turbines, how the immission of noise is influenced by propagation and why

psychoacoustics is important to interpret measurements and complaints correctly.

5.1. Search terms

Some thought was given to search terms since AM is not a widely recognised

phenomenon and might be reported under different names. The following English

search terms were used:

Amplitude Modulation

Aerodynamic Modulation

Modulated Noise

Wind Noise

Turbine Noise

Wind Farm Noise

Aero(-)acoustic Noise

Rotor Noise

Tonal Turbine

Fluctuating Noise

Swish.

Page 29 of 57


31/57


5.2. Information sources

The following journals were searched with the above terms.

Journals searched

1. Journal of the Acoustical Society of America

2. Journal of Sound and Vibration

3. Journal of Low Frequency Noise, Vibration and Active Control

4. Applied Acoustics

5. Technical Acoustics

6. Acoustical Science and Technology (Formerly Acoustical Society of Japan)

7. Journal of Vibration and Acoustics

8. Noise and Vibration Worldwide

9. Renewable Energy

10.Wind Energy

11.Renewable and Sustainable Energy Reviews

12.Energy

13.Journal of wind engineering and industrial aerodynamics

14.Journal of solar energy engineering

15.Applied energy

16.All relevant IEEE Journals

17.Some papers which appear as abstracts only in JASA

18.International Journal of Energy Research,

19.American Institute of Aeronautics and Astronautics Journal

20.Transactions of the ASME: Journal of Dynamic Systems Measurement and

Control and Renewable Energy.

Conference proceedings search

1. Wind Turbine Noise 2005

2. Internoise 1996

3. Euronoise 2003 & 20064. BWEA Wind Energy Conferences 1997 - 2005

Page 30 of 57


32/57


5. Forum Acusticum 2003-2006

Other Sources Found

1. Government Reports Scotland, Denmark, Wales, Sweden etc

2. Books

3. General public reports

Because there have been major wind energy developments in Germany in the recent

past, articles in German using German search terms were also searched. There are

no likely journal sources in German, and the only likely conference (DAGA) only

publishes abstracts rather than full papers. Therefore the German language search

focussed on the internet. Only two papers have been found in this way which were

from the mid 1990s and in English.

5.3. Government reports and regulations

The current British guidelines and regulations are detailed in PPS2 (2004) for

England, PAN45 (2002) for Scotland and TAN8 (2005) for Wales which can all be

traced back to ETSU-R-97 (1997). The general role of these is to define acceptable

noise levels for the protection of the population, to introduce consistent procedures for

establishing and rating noise levels and to minimise noise levels from wind turbines.

These documents were thoroughly reviewed in Hayes (2006).

Separate noise limits are applied for daytime and night-time. ETSU-R-97 indicates

that the purpose of these different noise limits is that for night-time periods, the

emphasis is on the prevention of sleep disturbance, whereas, the daytime noise limits

are to protect the amenity value of the area and of a property in particular.

The noise limits take the form of a fixed level for periods when background LA90

noise levels are very low and a margin above the background once background noise

levels increased due to wind effects. The daytime noise criterion is defined as the

greater of 35-40 dB LA90 or background + 5 dB and for the night-time period thegreater of 43 dB LA90 or background + 5 dB.

Page 31 of 57


33/57


ETSU-R-97 has summarised the issue of blade swish/AM as follows: The noise

levels recommended in this report take into account the character of noise described

as blade swish. Given that all wind turbines exhibit blade swish to a certain extent we

feel this is a common-sense approach given the current level of knowledge. The

character of blade swish is described as irregular enough to attract attention, to a

degree, turbine-dependent and dependent upon the position of the observer.

Since the publication of ETSU-R-97 a further review of European policies regarding

wind turbine noise has been carried out by Pedersen (2003). Of particular interest are

German legislation from 1999 and Dutch Legislation from 2001 as these were not

reviewed in ETSU-R-97. In German legislation different noise limits are applied to

different areas. The areas are industrial, mixed, residential and sensitive (for example,

hospitals or health resorts). In Dutch legislation wind turbine noise limits are based

upon a wind speed dependent curve which starts at 40dB (A) at night time, 50 dB (A)

during the day and 45 dB (A) during the evening. None of these regulations makes

any special allowances for AM.

5.4. Generation of wind turbine noise

There is a large body of literature relevant to noise generation by wind turbines. For

example, there are many publications concerning noise from aerofoils, propellers, fans

and helicopters which may be relevant. Fortunately, the fundamental issues for

windturbines have been summarised in two text books, and most of the following few

paragraphs are a distillation from these references.

Noise sources on wind turbines fall into two main categories: aerodynamic and

mechanical. Mechanical noise sources are primarily connected with the electrical

generation parts of the turbine, the gear box and the generator. These sources are

generally located in the nacelle. The character of the noise generated is similar to that

from other types of rotating machinery, and often includes an audible tone or note that

might be described as a whine or hum. Noise with a pronounced tonal feature isgenerally recognised to be more annoying than noise of a comparable level but with

Page 32 of 57


34/57


neutral character. This subjective effect is recognised in the ETSU guidelines where a

penalty is added which is related to the audibility of the tone. Mechanical noise was

audible from early turbine designs. On modern designs the problem has been almost

completely eliminated, although it may arise temporarily if there is a mechanical fault,

such as worn bearings within the gear box/generator, worn teeth within the gear box,

or misalignment of the generator drive shaft. This discussion of mechanical noise does

not shed light on the causes of AM noise, but it does help to interpret some of the

complaint statistics, many of which were from early turbine designs or due to a fault

condition.

Apart from occasional fault conditions creating mechanical noise, the dominant noise

generating mechanisms on modern turbines are aerodynamic. Aerodynamic sound is

generated by pressure variations within the air which fluctuate at acoustic frequencies,

i.e. between about 20 and 20,000 times per second. In wind turbines such fluctuating

pressure is caused by flow turbulence. If hypothetically the flow of air into, and over

the blades could be made completely smooth the major noise sources would

disappear. In practice, the flow is generally turbulent and so some noise generation is

inevitable.

First, air flowing into the blades due to wind is not completely smooth, but contains

turbulent eddies of a range of sizes. These eddies are ingested into the blade region

and generate inflow turbulence noise as the blades chop through them. Second, as air

flows over the surface of the blades, turbulence is generated close to the surface in the

so called boundary layer. This boundary layer turbulence generates noise, particularly

when it interacts with the trailing edge of the blade, which is therefore known as

trailing edge noise. This is often the principal noise generating mechanism on wind

turbines. Other types of turbulence are the vortices shed from the tip which generates

tip noise or from the trailing edge of the blade. Trailing edge vortices are stronger

for blunt trailing edges and the associated noise is therefore called blunt trailing edge

noise. The above four mechanisms, inflow turbulence, trailing edge noise, tip noise

and blunt trailing edge noise, account for the majority of the noise from wind turbines,

and on modern design blunt trailing edge noise is not a significant effect.

Page 33 of 57


35/57


Other types of turbulence may also generate noise, but can be avoided. A condition

known as stall may occur and indeed is used to regulate rotational speed and power

generation for some designs. This can generate noise up to 10dB higher than without

stall. However, manufacturers are increasingly tending to move away from stall-

regulated machines, particularly for machines of higher power, one of the main

reasons being the higher noise levels they generate. Another possible cause of noise is

flow over imperfections in the blade surface, for example damage due to holes in

blades has been known to cause strongly noticeable tones. For large wind turbines

with good manufacturing quality control, such imperfections would be considered a

fault condition.

The frequency of the noise generated depends on the size of the turbulent eddies;

broadly speaking large eddies produce low frequency noise and small eddies generate

higher frequencies. Mostly, the character of aerodynamic noise is broad band, i.e. it

does not contain a distinguishable note or tone, but is of more random character such

as noise from rustling leaves, waterfalls, rain etc. The dominant character of the

combined aerodynamic noise as described above is therefore a swish, which is

familiar to most people who have stood near to a large wind turbine. Blade swish is

not completely steady, but is modulated (fluctuates) at the rate at which the blades

pass a fixed point, i.e. there is a cycle of increased and then reduced level which

typically at the blade passing frequency of around once per second. In the majority of

installations the modulation depth is a few dB which is subjectively acceptable. It is

not clear why in some situations the modulation depth increases to the point where it

becomes subjectively unacceptable, and therefore potentially annoying. In early wind

turbine designs, where the rotor was positioned downwind of the tower, a pronounced

thump was caused as each blade passed through the wake shed from the tower.

However, this effect is eliminated completely for the upwind rotor designs found on

all new windfarm developments since the large scale uptake of wind energy

development in the UK, and does not therefore explain the occurrence of AM.

It seems likely that AM is due to fluctuation in the strength of some of the above

mechanisms rather than to some completely new mechanism. Aerodynamic noise

generation depends primarily on the rotor tip speed, but there is also some dependence

on wind speed. Therefore, if wind speed is not even across the rotor plane then some

Page 34 of 57


36/57


fluctuation in level can be expected as the blade turns. Van den Berg has investigated

this possibility and has postulated that in stable atmospheric conditions the difference

in wind speed between the top and bottom of the rotor is relatively high. Therefore,

the wind speed seen by a blade varies cyclically, which in turn may cause the noise

level to vary once per revolution. This work is discussed in more detail in the next

section.

Sound generation by turbulence is a highly complicated phenomenon that is still not

completely understood despite a great deal of research into noise from fans, aircraft,

and propellers, and great advances since the 1950s. Mathematical models for

turbulence and turbulence generated noise have been developed but not to the stage

where reliable prediction is possible from the drawing board. The occurrence of AM

implies that fluctuations are occurring in the generating mechanisms, and the causes

of such variations are still less well understood. There are no existing models by

which AM can be predicted.

5.5. Propagation of wind turbine noise

In general noise propagation from wind turbines is determined by source directivity,

geometric spreading and atmospheric absorption, ground reflections and absorption,

meteorological effects and terrain complexity. The audibility of noise from wind

turbines is then determined by the ratio between turbine and background noise, the so-

called masking effect.

According to Oerlemans and Lopez (2005) AM noise from wind turbines is not

equally loud in all directions but is radiated primarily from the outer part of the

downward moving blade in the downward direction. This is the reason that AM noise

can often be heard underneath the turbine but not further away because at a

considerable distance the recipient is more in line with the horizontally propagated

weaker noise.

This directional sound is then spreading as it travels away from the turbine. The socalled geometric spreading decreases sound levels with increasing distance from any

Page 35 of 57


37/57


source. Atmospheric absorption also reduces the sound level with distance but is more

effective at high frequencies with the consequence that low frequency sound travels

further.

The main influence of meteorology is the change in sound speed profile due to wind

shear and temperature profiles which is well described in Wagner et al. (1996). Most

pronounced is the shadow zone upwind of the noise source and the increased noise

levels in the down wind direction. In extreme inversion conditions sound rays from

elevated sources can be bent back down to the ground downwind and create focussing

effects at considerable distances from the sound source resulting in slightly enhanced

noise immission relative to no-wind, no-inversion conditions (e.g. ISVR, 1991)..

To predict noise levels from wind turbines the wind speed is measured at 10 m height.

Using a typical wind profile for neutral atmospheric conditions the wind speed is then

extrapolated to hub height. In situations with large wind shear low wind speed near

the ground and high wind speed at hub height the background noise created for

example by vegetation noise close to the ground is small. The turbine blades

experience a higher wind speed resulting in unexpectedly high aerodynamic noise

levels. The lack of background noise then leaves the turbine noise more audible.

Several authors report this reduction of the masking effect (Harders and Albrecht,

2005, Sloth, 2005, Golec and Golec, 2005, van den Berg 2003, 2004, 2005a). Both

Klug (2005) and van den Berg (2003, 2004, 2005, 2005a, 2005b) find that high wind

shear situations occur frequently during the night time in so called stable atmospheric

conditions.

Because of its complex nature, the most debated meteorological influence on outdoor

sound propagation is that from atmospheric turbulence. In the noise shadow,

turbulence can lead to enhanced noise levels (Wilson 2000) compared to the high

levels of attenuation normally found upwind of a noise source. It also spreads the

sound more widely leading to a net-attenuation of highly directional noise (Salomons,

1994).

Many wind farms in countries like The Netherlands and Germany are situated in very

flat terrain with no obstacles to influence sound propagation between the turbines and

Page 36 of 57


38/57


the immission sites. In the UK many sites are a lot more complex than that; partly in

hilly or even mountainous terrain or in the middle of a forest. For these sites the sound

propagation becomes a lot more complicated too. Bolin (2005) dedicates an entire

publication to the systematic deployment of vegetation noise to mask turbine noise.

Apart from absorption and reflections from obstacles which attenuate and enhance

noise respectively, complex terrain also provides obstacles that act as a noise screen

(Jacobsen, 2005, Prospathopoulos and Voutsinas, 2006). This effect is frequency

dependent. If the size of the obstacle is a lot bigger than the wavelength it is an

effective screen. For very low frequencies and small objects this is not always the case

(Jacobsen, 2005).

One propagation phenomenon relating to AM noise is not yet well understood: In

some situations AM noise seems to travel and can be heard at a considerable distance

from the turbines. First explanation attempts have been published by van den Berg

(2003, 2004, 2005) and it appears likely that a combination of generation and

propagation mechanisms is responsible for this effect. Further studies are needed to

explain and predict the observed noise levels.

5.6. Amplitude Modulation of Wind Turbine

Aerodynamic Noise

Amplitude modulation of the noise associated with the operation of wind turbines was

identified within the Report The measurement of low frequency noise at three UK

wind farms: URN No: 06/1412 issued by the DTI in July 2006. Within the

conclusions of this report, the following were identified:

Infrasound associated with modern wind turbines is not a source which

will result in noise levels which may injurious to the health of a wind farm

neighbour;

Low frequency noise was measurable on a few occasions, but below the

existing permitted Night Time Noise Criterion;

Page 37 of 57


39/57


That the common cause of complaint was not associated with LFN, but the

occasional audible modulation of aerodynamic noise, especially at night.

The measurements undertaken and reported within the DTI Report identified periods

during wind turbine operation when noise associated with the operation of the wind

turbines varied in level at the blade passage frequency2

of the wind turbines. When

this variation in the noise associated with the operation of the wind turbines was noted

by neighbouring residents, it was described as a distant train or distant piling.

Measurements of the internal noise levels during these periods of wind farm operation

indicate that A-weighted noise levels are subject to amplitude modulation levels of

between 3 5 dB(A). Analysis of these periods using third octave band analysis

indicates that between 200 800 Hz, noise levels in specific frequency bands may

change between 8 10 dB. External measurements indicate that, for external A-

weighted changes in level of 3 4 dB(A), third octave band levels may change by

between 7 9 dB. Measurements reported for Wind Farm D (Table 1) have indicated

that third octave band levels when complaints were received before the

implementation of wind turbine control features, indicated level changes of 1215dB.

(All the above figures are ranges from peak to trough).

The finding that this modulation is concentrated between the frequency bands of 200

800 Hz is significant in that this is generally generated by the trailing edge of a wind

turbine blade. This has been identified as one of the main sources of aerodynamic

noise associated with the operation of wind turbines (Oerlemans and Lopez, 2005).

Trailing edge noise was identified by van den Berg (2006) as the most likely source of

the modulation which he observed at a wind farm on the Dutch/Germany border.

However, van den Berg postulates that the increased levels of modulation which he

observed were associated with the change in the aerodynamic environment seen by a

wind turbine blade as it passes in front of the wind turbine tower. He postulates that

when the turbine is at this point, a decrease in the wind speed associated with the

2 Blade passage frequency is the frequency at which the wind turbine blades pass a fixed point on the

wind turbine rotor, typically assumed to be the wind turbine support tower. For a wind turbineoperating at 30 rpm, with a three bladed rotor, this would equate to 90 blade passes a minute, or a blade

passage frequency of 1.5 Hz.

Page 38 of 57


40/57


presence of the tower will result in a change in the aerodynamic properties and

therefore the noise generated by the wind turbine. Van den Berg considers that this is

exacerbated by the increased atmospheric stability which occurs for his test site

during the night hours. With increased atmospheric stability, the wind speed seen at

the hub height of the wind turbine may be appreciably higher than that found at the

lowest part of the rotor path, which is also where the effect of flow disturbance around

the support tower is felt. Van den Bergs analysis indicated that the maximum change

in noise levels based upon these conditions was from 2 1 dB for neutral conditions

increasing to 4.8 1.7 dB for very stable atmospheric conditions. To achieve the

levels of modulation that van den Berg measured on the faade of neighbouring

dwellings, he suggests that multiple sources have added together to obtain increased

levels of modulation which phase in and out to give increased periods of

modulation of the aerodynamic noise.

However, measurement of the aerodynamic noise sources within Oerlemans and

Lopez (2005) indicate that the tower/blade interaction is a source which is at least 12

dB below that associated with the downward sweep of the turbine blade. It is

recognised that these measurements were undertaken in neutral atmospheric

conditions for a location relatively close to the wind turbine, however, there is almost

no indication of the presence of the wind turbine tower within the data. Therefore, the

effect of the tower on an upwind wind turbine noise emission character as postulated

by van den Berg is not borne out by measurements in the field.

Measurements of wind farm noise at sites in the UK indicate that where a wind farm

has periods of increased AM, these are not necessarily related to periods of high wind

shear. Wind Farm Site D, for example (see Table 1 above), is subject to periods of

increased amplitude modulation when the wind is blowing from the east. An easterly

wind does not increase the stability of the atmosphere seen by the wind turbines as

this noise is associated with day and night-time operations, however, topographical

effects result in some wind turbines being unsure as to the wind direction. This is

caused by the wind turbine wind vane being influenced by the wind direction at the

hub height of the rotor but the wind direction at the lower arc of the rotor may be from

a different direction. When specific wind turbines are stopped from operating the

modulation all but disappears. Wind speed measurements from tall free-standing

Page 39 of 57


41/57


anemometer masts located atop hills indicate that there is little change in atmospheric

stability between the day and night-time periods. However, locations in flat

landscapes do indicate an increase in the wind shear which is associated with stable

atmospheric conditionsbut not to the level described by van den Berg (2006).

On the basis of the above discussion, it is clear that the observed effects of amplitude

modulation which is sufficient in level to initiate complaints associated with this

acoustic feature is still not understood in sufficient detail to provide design guidelines

to minimise the potential for such a feature and is also still subject to debate.

5.7. Psychoacoustics

There are two sets of psychoacoustic related literature which are of interest when

considering the problem of AM noise emissions from wind farms; those which

examine the effect of modulated noise on listeners and those which examine

psychoacoustic phenomena directly caused by wind farms.

The perception of loudness is described in Moore (2004). The perceived loudness of a

sound is found to be dependent upon intensity and frequency content. It is also

different for different listeners. For wideband noise the smallest detectable amplitude

change amounts to 0.5-1 dB depending upon the sensitivity of the listener.

The effect of amplitude modulation within diesel locomotive noise with respect to the

annoyance it causes for humans is examined in Kanteralis and Walker (1988). It was

found through subjective tests that the noise from a diesel locomotive type was rated

as being annoying compared to an electric type locomotive despite both having

similar low frequency content. The testing was carried out in a situation which

replicates the listening conditions of a typical living room. The difference in

annoyance was found to be related to a pulsing effect or amplitude modulation in the

diesel engine case that occurred at the firing frequency of the diesel engine. The

authors comment that if the modulation frequency is increased above 12 Hz the

annoyance is decreased. However, as typical wind turbine blade passing frequencies

Page 40 of 57


42/57


are of the order of 1Hz this does not give much insight into the situation for wind

turbine noise.

In a paper on the annoyance and known psychoacoustic effects for wind turbine noise

Persson-Waye and Ohstrom (2002) report how the character of the noise from wind

turbines built by five different companies is perceived by a small number of listeners.

They conclude that levels of annoyance cannot be explained by the psychoacoustic

parameters loudness, sharpness, fluctuation strength and modulation. Several studies,

including Pedersen (2003), mention that although there is correlation between

annoyance levels and sound levels, the annoyance is also influenced by visual factors

and attitude to wind turbines.

Three publications by Pedersen and Persson-Waye (2004, 2005,

Documents

Aerodynamic Modulation Salford Uni